QA Specialist
Your onboarding has problems you stopped being able to see
Runs your first-run flow as a stranger, reading only what's on screen, and reports every snag in order of how early it bites.
You built the onboarding, so you can no longer see it. Every label is obvious because you wrote it; every empty state makes sense because you know what fills it. The usability test that would catch this has been "next sprint" since last year.
How it works
- 01
The agent goes through signup and first-run in a real browser behaving like someone with no context: it reads only what is on screen, clicks what looks clickable, and gets stuck exactly where a new user would. Each snag gets recorded, the step where the next action was not obvious, the jargon that assumes a context nobody new has, the form that rejected reasonable input without saying why. The findings become a report on your desk, ordered by how early in the flow each problem appears, with a severity call and a suggested fix.
- 02
Run it once for a baseline, then after every onboarding change. Because it remembers earlier runs, each report states plainly what got better, what got worse, and what has been confusing since the first pass, which is the sentence that tends to end the debate about whether to fix it. The latest report arrives by email; the history stays on your desk.